PAIRED ASSOCIATE LEARNING WITH DIFFERENTIAL REWARDS.
Technical rept. no. 66,
STANFORD UNIV CALIF INST FOR MATHEMATICAL STUDIES IN THE SOCIAL SCIENCES
Pagination or Media Count:
An attempt was made to separate the reward and information values associated with trial outcomes in a paired-associate situation. The anticipation procedure was used with a 25-item list. A trial began with a single letter appearing on a screen then the subject made one of two available responses and received one of a unique pair of point values associated with that item. The number of points received over a series of trials was directly related to the subjects monetary payoff for participation. Two independent groups of 24 subjects were run 1 Anoncorrection group was shown only the point value associated with the response made on a trial. 2 The correction group was always shown the points corresponding to both responses. Learning curves in terms of correct responses choices of the more favorable alternative per trial for the correction group exhibited no significant differences among the 25 payoff combinations studied. For the noncorrection group both rates of learning and asymptotic proportions of correct responding were directly related to the difference between the two payoffs associated with an item.